On Friday, September 16, 2011 09:58:59 you wrote:
> Dear FSL experts,
>
> For my resting state fMRI study, MELODIC ICA and dual regression yielded 20
> components. Now, I would appreciate it if someone would answer the
> following questions.
>
> First, how do I find out which component is (or is part of) which network
> (e.g. executive control network)? Do I do it visually by just comparing
> the (1) MELODIC output image of each component with the (2) pictures of
> the network included in other papers? Or is there a more reliable way to
> do this?
I basically just do it visually, like you said. If the component you're
looking for is associated with a task then it might also help to look as the
concordance between the putative component's time course and the task time
course.
> Second, after identifying the networks, do I jump straight into
> interrogating (i.e. extracting cluster size, cluster peaks, etc.) the
> chosen output images instead of running further processing or analyses?
Depends on what hypothesis you want to test. You should probably familiarize
yourself with dual regression if you aren't aware of it already.
http://www.fmrib.ox.ac.uk/analysis/dualreg/
> Many thanks,
>
> Kie Woo
|